## Electronic Theses and Dissertations

## Date

2021

## Document Type

Dissertation

## Degree Name

Doctor of Philosophy

## Department

Mathematical Sciences

## Committee Chair

Lih-Yuan Deng

## Committee Member

Dale Bowman

## Committee Member

Lih-Yuan Deng

## Committee Member

Dale Bowman

## Committee Member

Su Chen

## Committee Member

Ching-Chi Yang

## Abstract

Pseudo-random numbers (PRNs) are the basis for almost any statistical simulation and thisdepends largely on the quality of the pseudo-random number generator(PRNG) used. In this study, we used some results from number theory to propose an efficient method to accelerate the computer search of super-order maximum period multiple recursive generators (MRGs). We conduct efficient computer searches and successfully found prime modulus p, and the associated order k; (k = 40751; k = 50551; k = 50873) such that R(k; p) is a prime. Using these values of ks, together with the generalized Mersenne prime algorithm, we found and listed many efficient, portable, and super-order MRGs with period lengths of approximately 10e 380278.1;10e 471730.6; and 10e 474729.3. In other words, using the generalized Mersenne prime algorithm, we extended some known results of some efficient, portable, and maximum period MRGs. In particular, the DX/DL/DS/DT large order generators are extended to super-order generators.For r k, super-order generators in MRG(k,p) are quite close to an ideal generator. Forr > k; the r-dimensional points lie on a relatively small family of equidistant parallel hyperplanesin a high dimensional space. The goodness of these generators depend largely on the distance between these hyperplanes. For LCGs, MRGs, and other generators with lattice structures, the spectral test, which is a theoretical test that gives some measure of uniformity greater than the order k of the MRG, is the most perfect figure of merit. A drawback of the spectral test is its computational complexity. We used a simple and intuitive method that employs the LLL algorithm, to calculate the spectral test. Using this method, we extended the search for better DX-k-s-t farther than the known value of k = 25013: In particular, we searched and listed better super-order DX-k-s-t generators for k = 40751; k = 50551, and k = 50873.Finally, we examined, another special class of MRGs with many nonzero terms known as the DW-k generator. The DW-k generators iteration can be implemented efficiently and in parallel, using a k-th order matrix congruential generator (MCG) sharing the same characteristic polynomial. We extended some known results, by searching for super-order DW-k generators, using our super large k values that we obtained in this study. Using extensive computer searches, we found and listed some super-order, maximum period DW(k; A, B, C, p = 2e 31 - v) generators.

## Library Comment

Dissertation or thesis originally submitted to ProQuest

## Notes

embargoed

## Recommended Citation

Pasiah, Kenneth Bobvah, "EFFICIENT COMPUTER SEARCH FOR MULTIPLE RECURSIVE GENERATORS" (2021). *Electronic Theses and Dissertations*. 2942.

https://digitalcommons.memphis.edu/etd/2942

## Comments

Data is provided by the student.